config.json

{
  "architectures": [
    "SMBVisionModel"
  ],
  "auto_map": {
    "AutoConfig": "configuration_smb_vision.SMBVisionModelConfig",
    "AutoModel": "modeling_smb_vision.SMBVisionModel"
  },
  "dtype": "bfloat16",
  "hidden_size": 1152,
  "masking_ratio": 0.65,
  "model_type": "smb_vision_model",
  "predictor_config": {
    "depth": 12,
    "dtype": "bfloat16",
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 512,
    "in_channels": 1,
    "in_hidden_size": 1152,
    "initializer_range": 0.02,
    "intermediate_size": 1536,
    "model_type": "smb_vision_predictor",
    "num_heads": 16
  },
  "transformers_version": "4.57.3",
  "use_cache": true,
  "vision_config": {
    "deepstack_visual_indexes": [
      8,
      16,
      24
    ],
    "depth": 27,
    "dtype": "bfloat16",
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 1152,
    "in_channels": 1,
    "initializer_range": 0.02,
    "intermediate_size": 4304,
    "model_type": "smb_vision_encoder",
    "num_heads": 16,
    "num_position_embeddings": 2304,
    "out_hidden_size": 2048,
    "patch_size": 16,
    "spatial_merge_size": 2,
    "temporal_patch_size": 16
  }
}